Heaters

How does a person go about designing a heater? I want to heat the voltage reference I built years ago. I looked at PTC heaters but they seem too powerful and also costly, so I wondered about winding up some nichrome or kanthal or similar wire to do the job, but I have no clue how to go about choosing suitable wire or estimating how powerful the heater actually needs to be.
I'm thinking in terms of the reference being inside a metal box with the heater wire wrapped round it, sensor inside, the whole thing inside another box. Running about 35 degrees C perhaps?

A small, incandescent light bulb, with a cheap PID controller, and a small thermocouple.

https://www.wish.com/product/5aaf5e...MIkJqlocT75gIVTdbACh3NbgjUEAQYAiABEgLUnPD_BwE

A 1W (or whatever) ceramic resistor as the heating element?

Search for data on the National Semiconductor LM399 (LM199, LM299). It is a voltage reference and heater inside a semiconductor package. I thought it had a temperature control circuit inside, but this datasheet does not mention one.


ak

throbscottle said:

reference being inside a metal box with the heater wire wrapped round it

Click to expand...

How big a box?
If it's only for a single component or small assembly, can you use a commercial crystal oven?

I've used these for quite a few things in the past, they used to be quite common & there are a few on ebay:
**broken link removed**
**broken link removed**

There are other styles as well:
**broken link removed**

**broken link removed**

Beware of old crystal ovens, in many of them the temperature control is a simple bi-metal strip which gives ON/OFF control.
As a result the temperature (and hence the frequency) cycles with the ON/OFF heating.

Some later types use a simple two transistor and two resistor and thermistor circuit is used.

JimB

I used to make ovens controlled crystal oscillators (OCXOs). I used TO126 transistors as the heaters. Transistors have the advantage that they are designed to transfer heat into a heatsink, so they can equally well heat a small oven. The transistor can be the control element for the current going through it.

I used a thermistor and an op-amp to control the temperature. There has to be a current limit to make sure that the current isn't to large, and that the majority of the heat is dissipated in the transistor, even when the oven is cold just after being turned on. There is a significant time delay due to the thermal lag, so the electronic gain has to be limited to make sure that the system doesn't oscillate.

Wow, thanks for all the responses!
PID controller just sems like massive overkill, though I like the incandescent bulb idea. I have plenty of 3v and 6v bulbs I can use!
Crystal ovens - too expensive, though they are very purrrrdddyy
LM399 - self heating, very nice, definitely something for future reference (pardon the pun). But I'm wanting to heat a pre-existing 5v reverence. Well an improved version of itself anyway.
Resistor/transistor as heaters - looks be the way to go. I wonder which is best?
The reference as it stands is battery powered and I'd like to retain that. Are there any efficiency trade-offs between light-bulb / resistor / transistor (or for that matter, custom heating element) methods? Though I know its all just heat and it goes where it goes. I suppose the bulb is worst from that perspective because part of it's output is wasted as light...

Here is the schematic of the crystal heater which I mentioned earlier:



I have no details of the thermistor, the manual where this was cribbed from just has an in-house part number for it.
A bit of experimentation may be called for.

throbscottle said:

The reference as it stands is battery powered and I'd like to retain that.

Click to expand...

There could be a problem.
The hotter you run the heater, the more energy you will use.
I guess you have to consider the highest ambient temperature that this thing will work in, and select the heater temperature to be a few degrees above that.
Also, good insulation is your friend when it comes to minimising energy usage.

JimB

The heaters for the oven oscillators that I used to make had a very similar circuit to the one that JimB posted.

It has the current limit that I mentioned, and the heat sink pad of the BD436 is the collector, which is ground, so it is likely that no insulation is needed between the metal oven and the transistor, which improves heat transfer. The circuit needs a PTC thermistor, which is less common than an NTC, and the thermistor power may be significant as there will always be 12 V or more across it.

Our circuit had an op-amp between the thermistor and the transistor to allow a low-power potentiometer to adjust the temperature. The op-amp also allowed the gain to be easily adjusted by the value of the feedback resistor around the op-amp. Too much gain led to oscillation of the temperature, to little meant that the oven temperature changed more than was desirable when the outside temperature changes.

It's worth remembering that with a simple circuit like that, or even one with an op-amp, there will be a temperature change in the oven when the outside temperature changes. The heat dissipation will be much more at low ambient temperatures, so the transistor will have to be driven harder at low ambient temperatures, so the thermistor will have to have a lower resistance. That means that the thermistor will be running somewhat colder at low ambient temperatures than it will at high ambient temperatures. That's not to say that the oven won't work, it just that it won't isolate totally from ambient temperature changes.

Those familiar with voltage regulators will be used to regulators where the output voltage hardly changes at all when the supply voltage changes. Temperature regulation circuits like the one shown are different. There is a significant change in oven temperature when the ambient changes.


The best performance from oven oscillators is achieved by adjusting the oven temperature to the temperature at which the crystal has zero temperature coefficient. As that temperature varies by a few degrees between crystals, even in one batch, temperature adjustment of the oven is needed.

We even made a circuit which used a microcontroller to find the best temperature for each one. The response time of the crystal to temperature changes was so large that adjustment took many hours.

Just from the datasheet, the reference seems to level out around 60 deg C, Although now I've looked at it properly, it's not very sensitive to temp changes anyway.
I'll have a play with the parts I've got anyway. Thanks for the circuit Jim - I'll use that as my starting point.

jim beat me to it, but i remember seeing in either Radio-Electronics magazine or QST magazine a crystal oven that used a transistor as a heater. and then the whole thing was enclosed with styrofoam and a metal cover (a shield box from an old military radio).

Google Roman Black's site.
He designed a somewhat simple circuit.

It took me a while to work out TR2 in that Xtal oven circuit is for current limiting. I've had a look at mr RB's heater now. Definitely nice and simple!!
I had another thought. If I use a mosfet as the heating element, doesn't that become self-llmiting to some extent? Is there a reason not to try it?

The MOSFET will be self-limiting, but the current will be different for different start-up temperatures, and may not be repeatable for different MOSFETS.

RB's heater is very simple, but it needs to handle heat from both resistors and from the transistor.

A little bit of a deviation from the current discussion...

In another discussion group someone proposed the use of a Peltier device as a means of stabilising temperature for a frequency determining element.

One of the problems when using a simple heater is that the controlled temperature must be higher that the highest expected ambient temperature.

But when using a Peltier device, the controlled temperature could be anywhere withing reason, say 20degC.
The Peltier can both heat and cool, just reverse the direction of current flow.
Thus the possibility exists to provide temperature control which is more energy efficient, but at the cost of greater complexity for a control system and greater financial outlay for the necessary hardware.

JimB

The other issue with Peltier cells is the need to get rid of heat when cooling. If you are trying to cool electronics by a 30 or 40 °C, you may need to get rid of a lot of heat. On the other hand, an oven will always work better with more insulation.

Ah, so that's why people don't use mosfets for this then! Peltier a bit out of my league atm. Fascinating things though.
I have at least one to92 darlington. Roman's 5v circuit dissipates 250mw max in the resistors. Would be very slow to get my reference warmed up I think! So I need a gruntier version.
I received some lm321's in the post the other day (because they were cheap). Is it worth going the op-amp route?